cryptospore / rpms / qemu-kvm

Forked from rpms/qemu-kvm 2 years ago
Clone

Blame SOURCES/kvm-migration-move-some-code-to-ram_save_host_page.patch

26ba25
From 35790a371aa43b6cc357bc78ee07dd20db16ec4b Mon Sep 17 00:00:00 2001
26ba25
From: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
26ba25
Date: Wed, 1 Aug 2018 13:55:10 +0100
26ba25
Subject: [PATCH 06/21] migration: move some code to ram_save_host_page
26ba25
26ba25
RH-Author: Dr. David Alan Gilbert <dgilbert@redhat.com>
26ba25
Message-id: <20180801135522.11658-7-dgilbert@redhat.com>
26ba25
Patchwork-id: 81570
26ba25
O-Subject: [qemu-kvm RHEL8/virt212 PATCH 06/18] migration: move some code to ram_save_host_page
26ba25
Bugzilla: 1594384
26ba25
RH-Acked-by: Peter Xu <peterx@redhat.com>
26ba25
RH-Acked-by: John Snow <jsnow@redhat.com>
26ba25
RH-Acked-by: Juan Quintela <quintela@redhat.com>
26ba25
26ba25
From: Xiao Guangrong <xiaoguangrong@tencent.com>
26ba25
26ba25
Move some code from ram_save_target_page() to ram_save_host_page()
26ba25
to make it be more readable for latter patches that dramatically
26ba25
clean ram_save_target_page() up
26ba25
26ba25
Reviewed-by: Peter Xu <peterx@redhat.com>
26ba25
Signed-off-by: Xiao Guangrong <xiaoguangrong@tencent.com>
26ba25
Message-Id: <20180330075128.26919-7-xiaoguangrong@tencent.com>
26ba25
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
26ba25
(cherry picked from commit 1faa5665c0f1df2eff291454a3a85625a3bc93dd)
26ba25
Signed-off-by: Danilo C. L. de Paula <ddepaula@redhat.com>
26ba25
---
26ba25
 migration/ram.c | 43 +++++++++++++++++++------------------------
26ba25
 1 file changed, 19 insertions(+), 24 deletions(-)
26ba25
26ba25
diff --git a/migration/ram.c b/migration/ram.c
26ba25
index 8dc98a5..106fcf1 100644
26ba25
--- a/migration/ram.c
26ba25
+++ b/migration/ram.c
26ba25
@@ -1484,38 +1484,23 @@ err:
26ba25
  * Returns the number of pages written
26ba25
  *
26ba25
  * @rs: current RAM state
26ba25
- * @ms: current migration state
26ba25
  * @pss: data about the page we want to send
26ba25
  * @last_stage: if we are at the completion stage
26ba25
  */
26ba25
 static int ram_save_target_page(RAMState *rs, PageSearchStatus *pss,
26ba25
                                 bool last_stage)
26ba25
 {
26ba25
-    int res = 0;
26ba25
-
26ba25
-    /* Check the pages is dirty and if it is send it */
26ba25
-    if (migration_bitmap_clear_dirty(rs, pss->block, pss->page)) {
26ba25
-        /*
26ba25
-         * If xbzrle is on, stop using the data compression after first
26ba25
-         * round of migration even if compression is enabled. In theory,
26ba25
-         * xbzrle can do better than compression.
26ba25
-         */
26ba25
-        if (migrate_use_compression() &&
26ba25
-            (rs->ram_bulk_stage || !migrate_use_xbzrle())) {
26ba25
-            res = ram_save_compressed_page(rs, pss, last_stage);
26ba25
-        } else {
26ba25
-            res = ram_save_page(rs, pss, last_stage);
26ba25
-        }
26ba25
-
26ba25
-        if (res < 0) {
26ba25
-            return res;
26ba25
-        }
26ba25
-        if (pss->block->unsentmap) {
26ba25
-            clear_bit(pss->page, pss->block->unsentmap);
26ba25
-        }
26ba25
+    /*
26ba25
+     * If xbzrle is on, stop using the data compression after first
26ba25
+     * round of migration even if compression is enabled. In theory,
26ba25
+     * xbzrle can do better than compression.
26ba25
+     */
26ba25
+    if (migrate_use_compression() &&
26ba25
+        (rs->ram_bulk_stage || !migrate_use_xbzrle())) {
26ba25
+        return ram_save_compressed_page(rs, pss, last_stage);
26ba25
     }
26ba25
 
26ba25
-    return res;
26ba25
+    return ram_save_page(rs, pss, last_stage);
26ba25
 }
26ba25
 
26ba25
 /**
26ba25
@@ -1544,12 +1529,22 @@ static int ram_save_host_page(RAMState *rs, PageSearchStatus *pss,
26ba25
         qemu_ram_pagesize(pss->block) >> TARGET_PAGE_BITS;
26ba25
 
26ba25
     do {
26ba25
+        /* Check the pages is dirty and if it is send it */
26ba25
+        if (!migration_bitmap_clear_dirty(rs, pss->block, pss->page)) {
26ba25
+            pss->page++;
26ba25
+            continue;
26ba25
+        }
26ba25
+
26ba25
         tmppages = ram_save_target_page(rs, pss, last_stage);
26ba25
         if (tmppages < 0) {
26ba25
             return tmppages;
26ba25
         }
26ba25
 
26ba25
         pages += tmppages;
26ba25
+        if (pss->block->unsentmap) {
26ba25
+            clear_bit(pss->page, pss->block->unsentmap);
26ba25
+        }
26ba25
+
26ba25
         pss->page++;
26ba25
     } while ((pss->page & (pagesize_bits - 1)) &&
26ba25
              offset_in_ramblock(pss->block, pss->page << TARGET_PAGE_BITS));
26ba25
-- 
26ba25
1.8.3.1
26ba25